Multisensor joint tracking and identification using particle filter and Dempster-Shafer fusion

Simultaneous multi-target tracking and identification using multiple radar sensors is advantageous to offer more reliable real-time information for situation assessment, resource management and decision making, which is essentially a problem of joint tracking, association, identification and sensor fusion. This paper first presents a method to use the Rao-Blackwellised particle filter (RBPF) based approach to address the joint multitarget tracking, association and identification in presence of clutter using a single radar kinematic measurement. Using the particle filter as an association indicator, the data association is efficiently integrated into the RBPF frameworks. To achieve more robust and reliable performance, multi-sensor fusion is exploited. Dempster-Shafter (D-S) belief function is then incorporated into the RBPF framework under the transferable belief model (TBM) to provide a flexible fusion result. Computer simulations using the proposed schemes show reliable tracking and reasonable and correct target classification with great flexibility.